Part XI · Swing trading · Chapter 11H

Four swing setups

The core archetypes

1 · What a setup specifies

The framework supports four primary swing setups, or archetypes, for index and large-cap single names. Each is a formal pattern with measurable entry conditions, sizing rules and invalidation criteria. Together they cover what the Dealer Mechanics Lab calls the high-probability operational surface for swing trading on horizons of 2 to 10 days. A fifth setup, pre-earnings momentum on single names, has Chapter 11I to itself, because its catalyst dynamics need the room.

Every setup answers the same questions: where its edge comes from, which regime it requires, what triggers it, and how to execute and size it. It also says in advance where it takes profit, when it gives up on time and what proves it wrong. Those last answers use units worth defining first.

R is the dollar loss you take if the stop is hit, so a profit target of 2R is a gain of twice that. A trade's R-multiple is its profit or loss divided by R. A time stop closes a trade after a set number of trading days, whatever its profit or loss. Chapter 11N gives the reason: a setup that has not paid within its typical horizon has a low expected value from there on.

Sizing from the stop

"Risk 1% of the portfolio" means the dollar loss if the stop is hit, not the notional deployed (Chapter 13C). With a risk budget $B$ in dollars, a stop distance $d$ per unit of the underlying and a multiplier $m$, which is 1 for shares, the number of units is

where the brackets round down to a whole number. The notional is $n\,m\,S$, so each dollar of risk carries about $S/d$ dollars of notional: one over the stop's distance as a fraction of the price. For an option, $d$ is the fall in the option's own price at the stop, which depends on its delta (Chapter 5C).

The Dealer Mechanics Lab puts the notional of equity swing trades at typically 20 to 100 times the dollar risk, which corresponds to stops 1% to 5% from the price. The tighter stops of the two mean-reversion setups below give more, 100 to 200 times at 0.5% to 1.0%. A long option bounds the notional by its premium.

2 · Setup 1: trend pullback and reclaim

The Trend Pullback, in full the Trend Pullback and Reclaim, buys a dip inside an uptrend.

Edge source. Mean reversion within a confirmed Stage 2 trend. In the Dealer Mechanics Lab's account, dealers with positive gamma exposure fade the pullback by buying at moving-average support, which gives the trade a mechanical tailwind (Chapter 8D).

The regime asks for more than the bull-tilt bucket of Chapter 11G, which starts at +10. It also asks for a narrower ADX window than the Stage 2 classifier of Chapter 11E, which needs only ADX above 18.

The trigger is RSI(2), the relative strength index over just two bars, or the internal bar strength (IBS). With $\bar{U}$ and $\bar{D}$ the Wilder averages of the gains and losses, RSI is $100 - 100/(1 + \bar{U}/\bar{D})$. Over two bars, Wilder smoothing gives each new change half the weight, so a couple of down days drive RSI(2) toward zero. IBS says where the day closed within its own range, with high $H$, low $L$ and close $C$:

How ATR, ADX and the normalized slope are computed

The true range of bar $t$ counts an overnight gap as part of its movement, using the previous close $C_{t-1}$:

$$\begin{aligned} \text{TR}_t = \max\big(&H_t - L_t,\ |H_t - C_{t-1}|, \\ &|L_t - C_{t-1}|\big) \end{aligned}$$ (11H.3)

The average true range (ATR) smooths it the Wilder way over 14 bars, $\text{ATR}_t = (13\,\text{ATR}_{t-1} + \text{TR}_t)/14$. Take a bar with a high of 5,950 and a low of 5,890 after a close of 5,960. Its true range is 70, the fall from the previous close to the low, not the 60 of the bar itself, and it lifts an ATR of 55 to $(13 \times 55 + 70)/14 = 56.07$.

The average directional index (ADX) measures how one-sided recent moves have been. The up-move $H_t - H_{t-1}$ counts as $+\text{DM}_t$ when it is positive and larger than the down-move $L_{t-1} - L_t$, and $+\text{DM}_t$ is zero otherwise. The down-move counts as $-\text{DM}_t$ by the mirror rule. Smoothing $\text{TR}$, $+\text{DM}$ and $-\text{DM}$ the Wilder way over 14 bars gives

$${+\text{DI}} = 100\,\frac{+\text{DM}}{\text{TR}}, \quad {-\text{DI}} = 100\,\frac{-\text{DM}}{\text{TR}}$$ (11H.4)

$$\text{DX} = 100\,\frac{|{+\text{DI}} - {-\text{DI}}|}{{+\text{DI}} + {-\text{DI}}}$$ (11H.5)

ADX is the Wilder average of DX over 14 bars. With smoothed values of 60 for TR, 18 for $+$DM and 6 for $-$DM, the indicators are 30 and 10, and DX is $100 \times 20/40 = 50$. An ADX of 22 then moves to $(13 \times 22 + 50)/14 = 24.0$, inside this setup's window.

Chapter 11E's normalized slope divides the 30-bar SMA's rise per bar over the last five bars by the ATR. With the SMA at 5,900 now and 5,880 five bars ago, and an ATR of 60, it is $(20/5)/60 = 0.067$, above +0.05, so Stage 2's slope condition holds.

Common failure mode. The setup fires with the composite between +10 and +30, a mediocre regime, and the price rolls over rather than bouncing. The mitigation is to respect the composite filter strictly.

3 · Setup 2: compression breakout

The Compression Breakout buys the first expansion out of a range that has been tightening.

Edge source. Range expansion after a contraction in volatility reflects supply that has been absorbed, the mechanism of the volatility contraction pattern (VCP) in Chapter 11D. In positive-GEX regimes, the Dealer Mechanics Lab adds, dealers short the upside hedge into the breakout. Because they are short calls above the range, they must buy as the price rises through them, which accelerates the continuation. Chapter 8D's regimes pull the other way on average, since positive aggregate GEX damps moves. The mechanism therefore rests on the calls dealers are short just above the range rather than on the aggregate sign.

Common failure mode. Fakeout breakouts on weak volume, which reverse within 2 days. The mitigation is to enforce the volume rule strictly. A breakout on volume below 100% of its average is statistically more often a fakeout than a continuation.

The figure draws the worked breakout with its whole plan. Drag the breakout volume below 130%, or the fill past 602.60, and the trade disappears. Then choose what the price does next, and watch which exit closes the trade.

breakout volume

145%

fill

602.50

then plan run fakeout stall

stop 596.00, R = 6.50

461 shares, $277,753 (93%)

target 614.00 (2R 615.50)

time stop: day 8

Take the breakout. Volume is 145% of its 20-day average, and the fill, 602.50, is within 0.1% of the pivot.

4 · Setup 3: undercut and reclaim

The Undercut and Reclaim buys a failed breakdown through a level that everyone is watching.

Edge source. Stop-loss orders cluster below well-known technical levels, such as the 200-day moving average or prior swing lows. When the price breaks the level, those stops trigger and accelerate the decline. When the move fails to follow through and reverses, the short sellers who opened on the breakdown are forced into short covering. The reclaim above the level signals that demand has overwhelmed supply.

Common failure mode. Several undercuts in a row, as the level becomes a magnet for further declines. The mitigation: if a setup undercuts a major level for a second time within 5 sessions, skip it, because the level is becoming a ceiling, not a floor.

5 · Setup 4: IBS and RSI mean reversion

IBS and RSI Mean Reversion buys a day that closes weak after a short slide.

Edge source. Combining intraday positioning, measured by IBS, with a multi-day oscillator, RSI(2), captures different aspects of selling pressure. IBS reflects today's close relative to today's range, while RSI(2) reflects a multi-day extreme of momentum. The Dealer Mechanics Lab argues that their confluence is genuine, not multicollinear.

This setup has a high win rate but a small payoff per trade, so it lives on its win rate. By Chapter 1C, the breakeven win rate is the average loss plus costs, divided by the sum of the average win and the average loss. If wins average 0.8R and losses 1R, the breakeven is $1/1.8 = 55.6\%$. Costs of 0.1R a trade raise it to $1.1/1.8 = 61.1\%$.

Common failure mode. Continuation lower if the regime is genuinely deteriorating, with the composite already weakening. The mitigation is to require a composite of +20 or more strictly.

Are IBS and RSI(2) two signals or one?

The Dealer Mechanics Lab is of two minds about this pair. This setup, and Chapter 13D's worked example of it, call the pair genuine confluence, because IBS measures a single day and RSI(2) several. Chapter 11C lists RSI(2) and IBS among indicators that all compute the same latent quantity: how stretched the price is against its recent range. It warns that counting them twice is multicollinearity, and Chapter 11N repeats the warning. Both positions are the source's own.

A day that closes at its low after a fall pushes both readings down at once, so the two are at least correlated. How correlated they are is an empirical question about your instrument. Before treating them as two votes, measure the correlation between IBS and RSI(2) over your own history, and ask how often one fires without the other.

6 · The four setups side by side

The four archetypes cover the conditional swing-trading opportunity surface across three different mechanisms:

Together they capture the high-probability surface for long swing trades in positive-regime environments. They do not cover short trades, which need Stage 4 or a negative regime. Nor do they cover long-volatility trades, which need an imminent event or extremely low IV. Multi-week positions driven by a thesis are out too, since they need a fundamental view and not just a technical setup. Chapter 11I takes up pre-earnings momentum, which does lie within this surface, and Chapter 11K the catalyst overlays.

7 · How strong the evidence is

For the Trend Pullback, the Dealer Mechanics Lab grades Connors' RSI(2) parameterization Tier B on the scale of Chapter 11B: practitioner work published in trade books, not in peer-reviewed journals. The underlying short-horizon mean reversion has academic support in Lo and MacKinlay (1990) and in Jegadeesh (1990), the reversal effect. The specific threshold, below 10, and the lookback are calibrated by practitioners. Treat the signal as practitioner-grade with an academic underlying, not as a Tier A result.

The Dealer Mechanics Lab grades it differently elsewhere. Chapter 13B's scale for the evidence factor lists "Connors RSI(2) mean reversion in trending markets" among its Tier 1 examples, scored 0.8 to 1.0, and scores this setup's evidence at 0.85. Both gradings are the Dealer Mechanics Lab's own, and only the cautious one comes with its reasons: the practitioner calibration described above.

The Edge Book files short-term reversal over 1 to 5 days under microstructure and flow: pay for providing liquidity after sharp moves (Chapter 1G). It is more doubtful than the Dealer Mechanics Lab that a retail account can capture short-horizon mean reversion at all, and Chapter 11B sets the two positions side by side.

Three of the setups require positive GEX, and two lean on dealer hedging for their mechanism. The Edge Book treats the sign of GEX as low-confidence context, worth 10% to 15% of one filter's weight. It ignores the sign when it disagrees with the VIX term structure or the realized-volatility filters (Chapter 10H). It rejects trading at specific GEX levels outright for retail accounts (Chapter 10I). On its reading, the GEX conditions in these tables deserve a small weight, and the stories about dealer flow a smaller one.

The other three setups cite no study. Chapter 11B places VCP detection among its Tier C examples, practitioner pattern language that you must formalize and test yourself.

8 · Check yourself